I think you did the BLTOUCH a disservice, Tom. While the other sensors perform equally well in any position, such as your horizontal position in the test fixture, the BLTOUCH is designed to work vertically. When put in the horizontal position, the sensor pin and magnet must overcome the additional friction of sliding along the bore of the guiding hole. This can lead to inconsistent measurements due to friction and grabbing. I consistently get average variation in readings around 0.015 mm
This!. The BLtouch is probably the best sensor on the market when used properly. It doesn't care what your build surface is and once your offset is determined you never have to worry about bed leveling or adjusting anything to do with the Z axis again. I routinely start a print and walk away, and expect it to come out perfect, and it does. The autoleveling is done once the bed is up to temperature, then the nozzle wipes across a brass brush, and quickly moves to the bed and starts printing. The BLTouch is an awesome bit of kit.
I think one of the things you have to remember is that the BLTouch is an all-in-one solution for bed leveling; it won’t interfere with your nozzle at all because of its retraction. That retraction most likely adds a significant wobble.
I would love to see this video updated to include Ultrasonic sensors, and name brand inductance sensors like those made by Osram up to say 16mm sensing distance. That being said, this video is so helpful. Thank you for making it.
ultrasonic too hard to work at small distance, basically time of flight is sooooo short electronics is not fast enough to process it. if you have at least 10cm distance then there is enough time to do calculation at reasonable precision yet you need to run higher frequencies. We tried to do auto focus device at 25mm distance even there we had to give up on ultrasonic sensors.
I want to address something I've been seeing floating around the internet regarding the inductive/capacitive probes. People keep saying you need a 12 V to 5V voltage divider on the output. This is partially wrong, and may, in fact, introduce more errors in your bed leveling. The sensors do absolutely need to be run from 12 volts to achieve their rated specs, but if you are careful to buy an NPN type sensor, you can simply plug the output directly into the Ramps input pin. This sensors are what is called "open collector drain" output. It means that it doesn't provide an output voltage when trigged, rather it pulls the signal provided to it to ground. The Arduino/Ramps board has internal 20 kohm pull up resistors on all of its pins (enabled in Configurations.h). This pull up provides 5 volts to the output pin of the sensor, when the sensor triggers, it pulls that 5 volt signal to ground. If you use a voltage divider, you might reduce the 5 volts enough to cause unreliable switching in the Arduino. PNP type = needs voltage divider. NPN type = do not use voltage divider.
When I was working on the auto-alignment at Type A Machines, I was very surprised at the unsolvable nature of the auto-alignment problem. There's essentially no way to get the desired flatness when combining the standard deviation of the sensor, with the rated flatness of aluminum, and the thermal expansion when heating to over 80C. You could conceivably do it with steel, but that has other challenges from the weight and cost of facing. Glass is really the only material we could find with below +/- 50 um flatness and dimensional stability when heated (we even managed to find some with +/- 0.3 um). Really the solution is to crowd source some magnetic glass, minimum order is 5 tons and works with induction.
Tom. You pretty much nailed this one....also you mentioned that capacitive sensors are thrown off by humidity...they are actually the instrument they use to sense humidity.
I never thought about using such sensors for measuring the bed level. At work I use mostly inductive ones but just for checking if there is an object or not. When I saw your video about the Prusa i3MK3 and its autoleveling I had doubts about the precision, but your video here has told me better. Thank you.
How are you only at 67K subscribers? Your videos are so good and the editing lighting etc is amazing! I wish you lots of success Thomas. I really enjoy every single video that you make.
Tom's videos are great and have amazing production value but how many 3D printer enthusiasts do you think are in the world? I imagine 67k people is a sizeable proportion of them.
Touché, but also lots of people that are interested in 3d printing don't have to be enthusiasts. Like you said though his production quality is probably some of the best on RUclips, and that is not an exageration. Many channels 10 hell even 100 times his size can learn a thing or two from him!
67k is quite a small subsection of 3d printing enthusiasts. Talk to Lulzbot and Ultimaker about their sale figures. Heck, Prusa and his company are shipping 3,000 printers a month, and that's only because that is the limit at what they can produce right now, giving new orders a 7 week lead time.
So this got me thinking what is the size of the global hobby 3d printing market? There's plenty of data on additive manufacturing market size but that's not people buying prusa i3 mkII's.
I know this video is over a year old at this point so I hope this is still relevant. He tested all of these sensors horizontally. The BLTouch specifically states NOT to use it that way. It needs to be hung vertically as it would be mounted on a printer. I've used a mechanical switch with manual leveling, a mechanical switch that flips up with a servo, capacitive sensors, inductive sensors, glass beds, PEI sheets, Zebra Plate from PRINTinZ, aluminum beds, and even tool steel beds. I've tried a total of 15 different sensors and 11 different beds in my few years of 3D printing. NONE of them were as accurate as simply using the BLTouch. When it is installed correctly, it just works flawlessly. Best of all it doesn't care one bit what surface you have, it will work on anything. I think it deserves a re-test.
It has almost no deviation on his tests, maybe 3 micron, which is next to nothing. He also states that it doesn't care about the surface material. So actually even in this horizontal test it comes out as (one of) the best choices regardless of where you want to print on.
+Thomas Sanladerer Great video! Aluminum tape definitely works though! I have used it multiple times, amd we also use it reliably at our Makerspace. There are different types of aluminum tape however. It looks like you are using the "metalized" type that is similar to Mylar. That stuff will not work. You need actual aluminum tape. The metalized type isn't very conductive. Actual Aluminum tape is basically just like aluminum foil with an adhesive backing. The only thing you have to watch out for is the adhesive. Alot of aluminum tape has a low temp adhesive and will start peeling up at around 50°C, but you can find some that has a high temp adhesive which is what we use under our PEI at the Maker Space.
You can tell by the video that it's the wrong kind of aluminum tape. Actual aluminum tape has a dull surface to it, and it is much thicker as well. Basically the tape you tested is just plastic tape with a small ammount of aluminum vapor deposited on the surface. Other than that, Great video!
Had minor problems with some Aluminum tapes and inductive sensors due to reduced sensing distances that I didn't like quite as well, found copper tape to be far better. But either works great. In fact on my setup(8mm inductive @12v) copper has better sensing distance than the aluminum heatbed(My printer crashes on autolevel if I leave the glass off and it tries to sense the aluminum sheet) Those here who use copper/aluminium tape under 3 or 4mm glass: What is your sensing distance? I put the tape on top and just don't print in the very corners as 1mm or so above the glass is too close for me with the tape under it.
I've been using a piece of heavy duty aluminum foil between my heated bed and glass for almost a year - with a 8mm inductive sensor it works great. It worked with a 4mm sensor as well - but was a bit too close for my tastes and Marlin's probe test seemed to show the 8mm was about the same for accuracy so I've stuck with it and been very happy.
That test machine was brilliant! I built an i3 clone following your guide and this sort of thing is really great for developing upgrades. Thanks for being so thoughtful in the design and execution of the testing.
If the tests were conducted as in the video , what do you make of Antcap's instructions regarding how the BLTouch must be mounted ? The manufacturer's website for the BLTouch clearly states this: if the sensor is mounted horizontally it WILL give wrong results. They don't equivocate. It is a crystal clear stipulation. It is underlined even . Would you care to comment ?
@@Traitorman.Con.14th.Sec3 except it is not just an electronic device . It is an electromechanical device and it is purpose designed to work mounted vertically . My kettle is a simple electronic device , it won't work as intended if I run it on its side or upside down.
@@Traitorman.Con.14th.Sec3 i take your point . But my inquiry is more to do with the method used to derive a conclusion , rather than the merits of this or any other device evaluated in the test. Manufacturer says this device X is designed and only works when used using method Y . Testing is done using method K , results turn out to be suboptimal . Why are the results suboptimal ? In truth we don't know ,. It may well be the product is unreliable but testing by using method K is not the way to find out. Over and out.
If I'd design a device like this small I'd surely factor the gravity whether I'd use a spring to push the pin out or not (as it happens I'm not a flat-earther, I "believe" in gravity). Probably the pin is pulled in by a coil (with the ferromagnetic pin acting as a core). Because it needs precision, the spring, if there is one, should be weak, but the trigger point will be more influenced by the sensor orientation. I guess you could make it switch just by shaking it in the hand. To be able to ignore the gravity direction (and to have repeatability no matter the sensor orientation), the spring should be stronger, but, in this case, the precision will go away, especially with microsteppping. Factor in the uneven friction, which can be significant for a such small device and the need to keep the price small and you'll have an answer. I don't own this small guy (BLTouch) and I didn't check the datasheet, but your observation is perfectly valid. And the BLTouch test is not valid.
Also the other mechanical sensor results aren't valid either because he's just testing the switch itself when to actually use it you'd need a mechanism to remove it to actually print.
one thing you're missing are FSRs, force sensing resistors. i use them on my delta which has a glass bed over an aluminum heat spreader. they seem to be very repeatable but only above 0.13mm deviation, any less than that and smoothieware's autocalibration gets upset
Thanks for your video, now I can comfortably purchase an induction sensor with the knowledge I will likely have to give up my glass bed. I appreciate all the testing you did!
Cool comparison. I was looking for a sensor to measure distance for my research project. So this helped me to understand the basics of the sensors. Thank you for that. Also is it possible to have the part list for your test setup? It will be really useful.
Great info, thanks Tom. I recently installed David Crocker’s IR sensor on my Prusa i3 box frame, which uses a 3mm glass bed and a red Mk2 PCB heated bed, inked black with a permanent marker in the sensor probing areas. After watching your video, I tweaked Marlin so that G30 would output the trigger height to 3 decimal places (1 micron resolution). I turned on my printer and made 46 measurements with G30 and got an s.d. of 1.4 microns. I noticed a trend in the data, presumably due to temperature drift somewhere in the system (probably the steppers warming up my crappy M5 threaded rods so the trigger point appeared to get lower with time). So I did another 34 measurements and the s.d. was 0.8 microns, but still with an obvious trend. I think these numbers are an order of magnitude better than you got with glass. Maybe it matters what is behind the glass? I should also mention there was a thin layer of glue stick on the bed under the probing area from the last print. Also I have slightly modded my sensor board to increase the trigger height from about 2.7 mm to 3.2 mm by tweaking the angle of the outer LED using a hot-air rework station. This is so that it would clear the bulldog clips I use to hold the glass down so I could still retain my full 200x200 build area.
That's because, most likely, your bed surface is suitable for Crocker's IR sensor. The testing surfaces and setup in the video is stated as not suitable in the documentation. If you are running a glass plate on top of bare aluminum it is recommended to paint the aluminum matte black. I wish this test was done with the proper setup.
Thanks for your videos. I have experienced difficulties with sensors. After a few consecutive prints, the levelling of the bed was not accurate, but was accurate again the day after. I now make the levelling when the bed is hot, but the problem still exists. I believe that the sensor close to the bed and the head warms up little by little, and the temperature of the sensor may have some effect of the measurement.
I have a question about the inductive sensors... How much 'material' do they need below and around them and at what point does it make a difference? For instance: If I trigger right on the edge of my aluminium plate it triggers lower then when probed in the middle of my plate. Does thickness play a huge role? Do they trigger higher when there is a bearing holder right below the sensing point? How thick does the aluminium need to be to not have that influence...
You measured mostly repeatability, and mostly ignored system accuracy due to heating/cooling. Excellent video, great charts. Should have had an external digital DTI measure the setups actual position, logging, and dwelled (paused) after contact on the probe. Then the DTI log would show where the moving piece actually was at probe contact, vs the start position when all was cold. Likewise, the triggering circuit for probe hit is critical. I did a lot of work on this, 8 years ago, on testing for lathe spindle index sensors, and we proved conclusively that sensors will give very fuzzy signals, that vary with temp/speed/luck. A sharp triggering circuit, and a sensor tuned to give a crisp response, will be about 100x more accurate on a range of speeds vs a typical probe "hit". I also saw that optical sensors will repeat to about 2 microns, with very simple basic cheap sensors.
Guden Thomas, wie immer: Sehr Informativ und Professionell! Ich danke dir für all deine Super Videos zum Thema, diese brachten mich in die Welt des 3D-Drucks! Weiter so, alles Gute und schöne Grüße aus dem Hessen, Otter
Would you please make a video explaining limitations of autoleveling? A lot of people seem to think it will compensate for misaligned axes (like when a dual motor Z axis loses motor sync and the X axis tilts).
I don't know where you're getting your sensors but I use 1 strip of aluminum tape under glass. Works like a charm. I used it as a retrofit onto the printrbot simple metal and it worked better than with the aluminum bed.
This is interesting. I see that microswitch w/o lever SD is about 1 um. Same as for inductive sensors. What is the point of using Bl touch or inductiv sensors if once can just put a microswitch button down on the print head and do autoleveling with it? but how to put it higher that the nozzle...
Another one to try is the piezo force sensor. Mount it in your bed support or (as I did) in your hot-end mount on the carriage and it registers contact of the nozzle to the bed. Zero probe offsets in X, Y and Z, change nozzle and all you have to do is run the auto-level. Super convenient. Precision Piezo do a good rig.
I feel like I've been in a time capsule. I got an early MakerBot CNC (#180) in 2009-ish, and was printing 3mm ABS on a raft on Kapton tape, using a .37mm layer height and no fans or heated beds. I woke up in 2021 receiving a Prusa Mini+ that prints directly on a magnetically attached and heated bed, using 50 micron layer height if I so desire, and supporting PLA, PETG and a plethora of other materials out of the box. In the old times, I tried using an optical mouse sensor for auto-leveling using focus. It actually worked with the sensor 1 mm from the bed, but I wanted something more like 10mm from the bed and never got the optics to work correctly. The idea was that the optical mouse sensor not only sensed the height, but also could read markings on the bed, or a printed sheet attached to it both for calibration precision and for communicating what to print. I never got to finish this project, but it seems like it still might be worth while: the inductive SuperPINDA doesn't even provide much headroom. Does anybody here have experience using optical focusing for Z height calibration?
For the mini-IR probe to be used on glass, the instructions say to put a black matte backing (black paper) behind it, or to paint the surface behind it matte black with high temp BBQ paint. A reflective surface behind it like aluminum will give the kind of results you saw in your tests.
Tony, see my clone of Davids mini IR, community.robo3d.com/index.php?threads/differential-ir-height-sensor-attemptin-clone-aint-sure-of-success-now-stage-10.6601/ ............Yes, you correct.....paint back of glass flat black.......Jimmy
1) 2) there must be distance for both of them. None of them will allow you to notice when the nozzle < 1cm. I guess there might be a workaround with offsets though when 5/10 centimeter or something like that.
Lidar is a process where an infared laser light creates a kind of radar around it. 3d Printers are only trying to measure distance in one direction. It is much simpliler and more practical to not use lidar. Ultrasonics are another radar type with help of a servo, Most radar like sensors are kind of inacurate but still acurate enoguh for their use cases like on a car or a plane. Where 3d printers are looking for accuracies of up to micrometers. Always learning, Aidan Villacampa
Tom, you really should continue wit the Myth-busting line. I have been experimenting with sensors I have to say that your conclusions corroborate my experience 100%. Given mechanical switches with lever , although less precise than inductive, can be used with borosilicate glass, I will stick to those. Awesome video, congrats.
This is awesome! Also fun fact: I use an aluminum mk3 heatbed with a 3/32 inch (2.38 mm) sheet of glass AND blue tape. The 8 mm inductive sensor triggers just fine for me thankfully.
Seeing as inductive probes can see through plastic/silicone, would it be sensible to put a silicone sock (or even a printed plastic cap) over the bottom of the inductive probe to reduce temperature effects? I'd be really interested to see if that helps.
Thanks for all that testing but now I want to know which probe is best for the Scoovo X9H. It has Kapton tape directly stuck on a glass bed with the silicon heating pad just under it. Board: RAMBo Firmware: Marlin
I have a unique idea for a sensor that might work better than any of these. I know it works much better than a Microswitch for sure at least....when I built it initially I didn't know what use it may be, but now it is all quite clear! Thank you Thomas! You inspire me!
Hello. Where can we read the results please? I am very interested by mecanical switch results (with and without lever) As I understand, at 7:24; mecanical without lever is more precise than BLTouch ? cool!!
"Alumininium" alone earned my subscription, whether it was intentional or just a typo. :D This was very informative, thanks for taking the time to do all the testing!
John Z Watch the video again if you didn't see what I was commenting on. I promise that no matter what country you're from, it's not generally spelled that way. Tom inserted a small item into the video that is internationally known as a "joke", and is intended to provoke a humor response in humans with at least average spelling and reading skills.
+John Z Seriously dude, look at what I wrote in my original comment, re-watch the video, and then try to be amazed at how dense you actually are. The word is intentionally misspelled both by Tom in the video, and by me above. There are extra letters in it. It doesn't say "aluminum" or "aluminium". In your efforts to claim Internet Superiority, you've completely missed the obvious.
I've just discovered your channel and love the friendly sound of your voice. I want to do a few small projects and I was weighing paying for prints vs buying my own printer. I found the new matter mod-t for 299 but wondered was there something better for beginners that maybe will help them as they transition to an intermediate level of skill without going a lot higher ( +$200 US) in price. In any case, I enjoy your videos. Thank you.
Great, informative videos. Thank you. I was wondering if you tested the BLTouch horizontally or vertically? I think that its accuracy is affected by gravity and would benefit from a vertical test. I agree that it is a good alternative for glass probing.
@@pellin-unleashthebiker3101 not when you're dealing with micron-level precision. I calibrate measurement equipment to this level as part of my job, and misusing equipment like this is a total non-starter if I want something to perform accurately.
@@TROPtastic did you look at the precision of the bltouch horizontally in this video? It is only some microns, so still perfect. So the conclusion is it really does not matter in this case. For something else it may, but not for a simple 3d printer.
Tom, if you retest the optical sensors, I would like to see it done with different color materials behind the glass (specifically black, white, red and peach- the common colors of heaters). Would also like to see how ambient IR light affects the repeatability, i.e. fluorescent lighting and sunlight. I had very mixed results with David Crocker's IR sensor during probing with sunlight coming in through the windows, for example, vs. probing in a dark room.
The IR sensor I make doesn't like bright sunlight (no clouds in front of the sun) shining directly on the bed and reflecting into the sensor. The current version 1.2 will flash the LED rapidly to warn you when it detects that. Otherwise, it isn't affected by ambient light. Incandescent light - especially halogen - affects some other types of IR sensor, but not mine. I'd also like to know what material was behind the glass when Tom did his tests. The reason that glass and other transparent materials can cause problems for optical sensors is that you get 2 or 3 reflections: one from the front, one from the back, and possibly a third from whatever is behind the glass. We want the one from the front to be stronger than the others, which is why we recommend a black surface behind the glass.
I liked this video a lot, I simply wish the sensor I use (David Crockers IR Sensor) had a slightly deeper/more in depth review. I am given more confidence to continue using it, however, will no longer calibrate it with the heated bed on. That is a variable I had not even thought twice of! I simply set the bed for 55° for my PolyMax PLA and calibrated it when it was done heating right before the print...Thank You for that!
Your videos are all excellent, but this one is superb! We spent so much effort learning this on our own... at Kikai Labs we use a capacitive sensor, take 16 measurements prior to starting each print job and level the bed automatically. It works great. We do calibrate the sensor first using three different bed temperatures, as it does matter as you pointed out. Interesting observation about the humidity, though... I used capacitive so that it would measure the glass surface, and we use a mirror instead of plain glass, and get better results. Also we found that the smallest the measuring distance, the more accurate it is, so we measure from 2mm away using a 12mm sensor.
I'd be interested in your review of the piezo electric z level sensor, it looks super interesting to me. It has a bed option and a hotend option, it's always there and has no XY offsets.
If you ever re-run this test I'd love to see the piezo sensors be included. I'm about to give one a try myself but curious to see your experience vs mine/others
Almost 2 years later now and I still adore my BLTouch. Even if it statistically or scientifically isn't the most accurate out there, as you say, it's 'more than good enough'. It's so rare to have a calibration fail.
Usually there is a spec sheet prescribing optimal materials and thickness with respect to frequency (typically a few mils for ferrous and 2-10x thicker for nonferrous). The relevant parameters are permeability and resistivity of the material. Ferrous materials and sensors generally can give higher resolution with larger sensors, and non-ferrous gives better resolution with very small sensors (even though typically smaller sensors give a shorter range), but there are universal sensors that have an equal correction factor for either. Generally ferrous sensors would probably work best in this application, and Generally target size should be 1-3 times the sensor diameter. Also, in terms of layered copper/alum tape, remember that you are isolating the layers electrically, so the induced eddy currents will be smaller (just like transformer laminations)… meaning you may need more layers of tape than you would expect. If I were adding a proximity sensor after the fact, I would get a medium-distance ferrous sensor, tape ~1mm thick steel disks to the bottom of the bed at the sensor points, fiddle with the distance, and call it a day.
This is amazing! Finally a useful application of the scientific method to get some hard data. It is so much more convincing than people's opinions. Love your videos!
How do i use a microswitch for the auto leveling? I mean where should i put it, so it is not interfering with the nozzle and mess up the print? It seems a good option for me. I have glass bed on top of an aluminum sheet... (DIY)
Wow love the detail and different perspectives covered in this video. I'm a sucker for this sort of data. If inductive sensors have their detection range halved, and there is a sheet of window glass (2.38mm thick) on top of the bed, doesn't that mean that both 2mm and 4mm inductive sensors are right out? I'd love to print ABS right on the aluminum but nothing sticks to it without a carefully maintained pile of glop (glue/slurry), where glass works for weeks with a single light hairspray application.
How does Marlin deal with the measurements from the probe? Will it compute a plane with minimum distance to each point, or does it compute a non planar surface to deal with deformations of the bed or bending/wear of the linear supports carrying the print head.
Very, very nice testing here! Thanks Thomas! I've been using a little microswitch (with lever) on a metal strip moved by a servo. It's working flawlessly for years now. (all metal printer, no 3d printed parts) Nice to see that this turns out to be a precise way indeed :-)
I like to print on glass and don't have auto leveling at the moment. The micro switch performed well but could you test it with a servo attached and see how it performs.
Informative as always. I use a LJ12A-4-Z/BY with an LM7805 VR on the output this gives a steady 5 volt trigger signal. If you increase the input voltage does it does it strengthen the sensing distance.
Necro, but thanks a billion for making this video. Would be interested to see what new sensors are out there, and learning more about the LIDAR from the Bambu Lab X1C
Great Video - seriously stellar quality and effort clearly went into this. The pinned top comment is exactly the question one is left wondering. Great job!
@@chloemcholoe3280 The glass (print bed) will be a flat piece of glass. However, it is then mounted inside a mechanical device. If one edge is mounted lower than the opposite edge then your first layer of 3d printing will not adhere well to the glass. The print bed must be leveled to ensure that the 3d print head glides over the whole print bed at the same height before a 3d print begins. Then the 3d printer will reliably 3d print the first layer and subsequent layers. This leveling can be done manually, with no automated sensors, or leveling can be done with automation using the kinds of sensors tested.
@@paull1316 yeah. my point was the reason I for these probes in my opinion is for fixing warpage. with glass you can just like live adjust the 4 corners as it's printing and it takes a few seconds really
@@chloemcholoe3280 - glass can be warped or bowed if not made correclty. Also, glass isn't the only height variable. its possible that the gantry rail that the hot end rides on can be bowed.. which may change the height of the nozzle as it moves back and forth.
When you tested the inductive sensors you didn't take in account the heterogeneity of an aluminium bed. While printing with my VORON I have experienced inconsistencies with inductive sensors and my PEI-painted Aluminium bed from Clever3D. Basically the trigger distance was consistent on the same spot (just like in your tests), but varied when probing a different spot... and you need to probe at least 3 distant spots. Today I switched to the VORON 1.5 X carriage, which uses the hotend itself (mounted on a mechanical microswitch) as a probe and the situation is vastly improved because a microswitch isn't affected by materials and temperature. EDIT: In theory this applies to IR probes too, as reflectivity may vary across the bed.
Hello Tom, I have a aluminum heated bed and a genuine e3dv6, and I was trying to use the autobed leveling making a "button" that closes when the brass nozzle touches the aluminum bed. Turns out that there is some kind of "coat" or resin or something in the surface of my heated bed so it doesn't conducts eletricity. Do you know what is this coating and why? And is this a good idea, use the nozzle itself?
Any chance to give me an idea, how to set up autoleveling in smoothieware, how to level the bed and what gcode should I put in start/end code? I have a 18mm capacitance proximity sensor on hotend carriage.
Very informative. I was wondering if you could repeat this experiment with the new MINDA probe from Prusa Mini. Many folks seem to be having a problem with leveling because of the variations between cold and heated bed.
Awesome test! Very methodical tests. I have the cap sense and have noticed the heat issues. I think I'll switch to inductive. I would like to try a glass bed over the aluminum lower plate. I wonder if the the longer range inductive sensor would be practical at detecting the aluminum glass combo.
How do you feel about auto levelling? I personally don't like the idea of the Z axis having to move during a layer print. If I'm worried I just use a raft, which will end up levelling before the actual model prints. What I really want to see is 3 servos to level the bed corners.
HenryLoenwind no idea what crazy setup you're envisioning but I guess it's not the one I was... I use servo in its loosest term: any motor hooked up to a potentiometer to increase positional accuracy, not as in your hobby king swivel arm 180° ones. As such, a small motor attached to the thumbscrew so it works as a linear actuator and a linear potentiometer hooked to that setup for positional feedback. Doesn't seem too ludicrous. You were suggesting using the nozzle to turn a thumbscrew? That grates against my engineering sensibilities quite badly.
+HenryLoenwind Oh man, the mental image you just gave me of a printer spinning its nozzle around a hooked screw to 'adjust' itself before it feels comfortable enough to start producing a part is hilarious to me, for some reason.
My team is actually working on a printer for the market with all best stuff available, and that includes 3 steppers for Z axis for REAL bed leveling, not some sort of lame workaround. ;) Fast hotbed (0-110C in 25 seconds), 300x300x450 build volume. Won't be cheap though. Good stuff is never cheap :)
I thought you needed an PNP sensor? (at least for a ender 3) Can you use either and set the high/low trigger in marlin firmware? Also, what voltage divider did you use?
Have you try the membrane film based zprobe (basically just like a membrane keyboard/contact switch)? I find it is the best zprobe for delta printer as it can be easily mounted directly under the nozzle to give the best on-points reading that a delta auto calibration needs (no zprobe offsets), without being affected by heavy trigger force that could potentially deform the belts/pivot points etc.
great video! cool rig, shows the power of rapid prototyping. been thinking about Auto leveling, I have a delta that uses pressure pads, would be great to see how well it stacks up, down side is Auto leveling is tapping printer head into the bed every time.
for my scratch built reprap delta we used a simple conductive type, running 1 wire to the nozzle, and another to the aluminium bed, reeally simple and easy, and no need to worry about effector tilt
You should try the aluminum tape trick with the optical sensors, I have a feeling the Sharp will benefit from it. Also, try doing tests with multiple sensors, like one for each corner of a bed vs a single sensor.
im using a glass bed and some aluminium foil in the corners underneath it, it manages to sense it but is it a good idea to continue using this construction?
What would the point be? He flat out told you it was done sideways to remove backlash from the equation and limit the results to just the component . Turning them upright wont change how the devices reacted mechanically one bit .
@@wind5250 and apparently you didn't read the instructions for the bill touch it supposed to be vertically standing not horizontally to work properlyso if you want to test something the way it was made to be tested and used that's the way you do it not for non-science a backlash there's no backlash when there's up and down
@@mathewphillips4185 not only did i read the instructions i own one. Laying it on it's side is just as valid as standing it up. The fact his results are within spec as well as repeatable , and comparable to other reviews more than demonstrates this .
@@wind5250 you read the instructions but apparently didn't follow them (and are misusing the equipment you own, but that's your choice). It's simple physics to demonstrate that there is a mechanical difference in how the probe behaves when it is vertical vs. horizontal: When vertical, gravity acts on the spring to pull it down and keep it in its rest position. When the probe hits the surface, it can cleanly move upward until it can trigger its Hall sensor. When horizontal, not only is gravity no longer aligning the spring correctly, but there is additional friction resisting the movement of the probe (thus causing the probe to move fractionally less for the same amount of deflection). When the probe contacts a surface horizontally, there is no force preventing the probe from deflecting sideways ten+ microns before triggering the Hall sensor, leading to a loss of precision. If using your equipment like this works for you, that's great, but it's objectively incorrect to say that orientation doesn't have an effect on the BLTouch and other precision equipment.
@@TROPtastic You took all that time to write a reply but obviously not enough to actually comprehend before hand . First of all i never said how i mounted the device let alone i mounted it sideways. What i said was it would make no difference to repeat the test upright because the results are within spec and repeatable which they are . You are literally arguing that the device will perform better than the manufactures tell you it will by turning it right side up. Anyone who actually uses this device knows they give invalid readings after time the only valid questions would be what version this is and was this one new or used .
Hi, I 'm not sure what's happening with my sensor which is the 8 mm inductive npn one. Before adding the Voltage divider I put in 12V across blue and brown and get a 12V signal on the black when I hold the sensor near metal and 0V when the metal is removed. When I add the Voltage divider as Tom did (double checked and resistors measured with multimeter to test) I get 3.6V signal on black when not near metal and 0V when I put sensor near metal. What have I done wrong? can I use this as a normally closed (nc) switch for bed levelling/tramming?
would it not be best to combine different sensor types in one device, for example inductive and contact, and measure the deviation of these for error analyses?
Great video. Love the rig you created. Is there a normal term for the slide you used that has to rails built in? I am thinking of using one on a 3D Printer I want to build from scratch
Hi, I use some lj12a3-4-z/bx on my printers and it triger on a glass plate with some almuminium foil sticked with rubber tape on it. Also supplied in 5V directly plug on my borads without any led or resistor.
Its been 4 years, will be cool to see a redo with current offerings
I'm officially finished with 3d/bl touches...they're both garbage, so yeah I'd like to see a 2021 sensor showdown myself 🙂
@@MAGA_Patriot2024 They really are garbage. for me after a long journey I came back to the source. end stop prob - klicky, the simpler the better
@@slimanus8m its been 5 years now
@@mynameisdex8785 Damn
@@mynameisdex8785 Ya....Now its official, just use Klicky or even Unklicky. it's simple cheap and accurate and cool
I think you did the BLTOUCH a disservice, Tom. While the other sensors perform equally well in any position, such as your horizontal position in the test fixture, the BLTOUCH is designed to work vertically. When put in the horizontal position, the sensor pin and magnet must overcome the additional friction of sliding along the bore of the guiding hole. This can lead to inconsistent measurements due to friction and grabbing. I consistently get average variation in
readings around 0.015 mm
This!. The BLtouch is probably the best sensor on the market when used properly. It doesn't care what your build surface is and once your offset is determined you never have to worry about bed leveling or adjusting anything to do with the Z axis again. I routinely start a print and walk away, and expect it to come out perfect, and it does. The autoleveling is done once the bed is up to temperature, then the nozzle wipes across a brass brush, and quickly moves to the bed and starts printing. The BLTouch is an awesome bit of kit.
0.015mm is a lot of deviation
This was awesome. Finally a proper comparsion between the different sensor types. Thanks a lot for your effort!
I think one of the things you have to remember is that the BLTouch is an all-in-one solution for bed leveling; it won’t interfere with your nozzle at all because of its retraction. That retraction most likely adds a significant wobble.
I would love to see this video updated to include Ultrasonic sensors, and name brand inductance sensors like those made by Osram up to say 16mm sensing distance. That being said, this video is so helpful. Thank you for making it.
ultrasonic too hard to work at small distance, basically time of flight is sooooo short electronics is not fast enough to process it. if you have at least 10cm distance then there is enough time to do calculation at reasonable precision yet you need to run higher frequencies. We tried to do auto focus device at 25mm distance even there we had to give up on ultrasonic sensors.
I want to address something I've been seeing floating around the internet regarding the inductive/capacitive probes. People keep saying you need a 12 V to 5V voltage divider on the output. This is partially wrong, and may, in fact, introduce more errors in your bed leveling. The sensors do absolutely need to be run from 12 volts to achieve their rated specs, but if you are careful to buy an NPN type sensor, you can simply plug the output directly into the Ramps input pin. This sensors are what is called "open collector drain" output. It means that it doesn't provide an output voltage when trigged, rather it pulls the signal provided to it to ground. The Arduino/Ramps board has internal 20 kohm pull up resistors on all of its pins (enabled in Configurations.h). This pull up provides 5 volts to the output pin of the sensor, when the sensor triggers, it pulls that 5 volt signal to ground. If you use a voltage divider, you might reduce the 5 volts enough to cause unreliable switching in the Arduino. PNP type = needs voltage divider. NPN type = do not use voltage divider.
Is this going to be the sensor version of Fillaween? ... Sensorween?
你內部人員幹嘛
When I was working on the auto-alignment at Type A Machines, I was very surprised at the unsolvable nature of the auto-alignment problem. There's essentially no way to get the desired flatness when combining the standard deviation of the sensor, with the rated flatness of aluminum, and the thermal expansion when heating to over 80C. You could conceivably do it with steel, but that has other challenges from the weight and cost of facing. Glass is really the only material we could find with below +/- 50 um flatness and dimensional stability when heated (we even managed to find some with +/- 0.3 um).
Really the solution is to crowd source some magnetic glass, minimum order is 5 tons and works with induction.
I'm game, magnetic glass ftw
just epoxy glue a steel sheet to a piece of glass
Tom. You pretty much nailed this one....also you mentioned that capacitive sensors are thrown off by humidity...they are actually the instrument they use to sense humidity.
I never thought about using such sensors for measuring the bed level. At work I use mostly inductive ones but just for checking if there is an object or not. When I saw your video about the Prusa i3MK3 and its autoleveling I had doubts about the precision, but your video here has told me better. Thank you.
Especially good description of what standard deviation means in this situation. Gut gemacht!
How are you only at 67K subscribers? Your videos are so good and the editing lighting etc is amazing! I wish you lots of success Thomas. I really enjoy every single video that you make.
He could get millions of subscribers from the FNL (GDR) which never learned english.
Tom's videos are great and have amazing production value but how many 3D printer enthusiasts do you think are in the world? I imagine 67k people is a sizeable proportion of them.
Touché, but also lots of people that are interested in 3d printing don't have to be enthusiasts. Like you said though his production quality is probably some of the best on RUclips, and that is not an exageration. Many channels 10 hell even 100 times his size can learn a thing or two from him!
67k is quite a small subsection of 3d printing enthusiasts. Talk to Lulzbot and Ultimaker about their sale figures. Heck, Prusa and his company are shipping 3,000 printers a month, and that's only because that is the limit at what they can produce right now, giving new orders a 7 week lead time.
So this got me thinking what is the size of the global hobby 3d printing market? There's plenty of data on additive manufacturing market size but that's not people buying prusa i3 mkII's.
I know this video is over a year old at this point so I hope this is still relevant.
He tested all of these sensors horizontally. The BLTouch specifically states NOT to use it that way. It needs to be hung vertically as it would be mounted on a printer.
I've used a mechanical switch with manual leveling, a mechanical switch that flips up with a servo, capacitive sensors, inductive sensors, glass beds, PEI sheets, Zebra Plate from PRINTinZ, aluminum beds, and even tool steel beds. I've tried a total of 15 different sensors and 11 different beds in my few years of 3D printing. NONE of them were as accurate as simply using the BLTouch. When it is installed correctly, it just works flawlessly. Best of all it doesn't care one bit what surface you have, it will work on anything.
I think it deserves a re-test.
It has almost no deviation on his tests, maybe 3 micron, which is next to nothing. He also states that it doesn't care about the surface material. So actually even in this horizontal test it comes out as (one of) the best choices regardless of where you want to print on.
Good video.
I have thought about switching to inductive sensor but now I just think I can keep the microswitch without the lever.
+Thomas Sanladerer Great video! Aluminum tape definitely works though! I have used it multiple times, amd we also use it reliably at our Makerspace. There are different types of aluminum tape however. It looks like you are using the "metalized" type that is similar to Mylar. That stuff will not work. You need actual aluminum tape. The metalized type isn't very conductive. Actual Aluminum tape is basically just like aluminum foil with an adhesive backing. The only thing you have to watch out for is the adhesive. Alot of aluminum tape has a low temp adhesive and will start peeling up at around 50°C, but you can find some that has a high temp adhesive which is what we use under our PEI at the Maker Space.
You can tell by the video that it's the wrong kind of aluminum tape. Actual aluminum tape has a dull surface to it, and it is much thicker as well. Basically the tape you tested is just plastic tape with a small ammount of aluminum vapor deposited on the surface. Other than that, Great video!
Agreed. I have had great results with aluminum tape on glass beds!
I've used copper tape in the past on glass, that worked pretty well as well.
Had minor problems with some Aluminum tapes and inductive sensors due to reduced sensing distances that I didn't like quite as well, found copper tape to be far better. But either works great.
In fact on my setup(8mm inductive @12v) copper has better sensing distance than the aluminum heatbed(My printer crashes on autolevel if I leave the glass off and it tries to sense the aluminum sheet)
Those here who use copper/aluminium tape under 3 or 4mm glass: What is your sensing distance? I put the tape on top and just don't print in the very corners as 1mm or so above the glass is too close for me with the tape under it.
I've been using a piece of heavy duty aluminum foil between my heated bed and glass for almost a year - with a 8mm inductive sensor it works great. It worked with a 4mm sensor as well - but was a bit too close for my tastes and Marlin's probe test seemed to show the 8mm was about the same for accuracy so I've stuck with it and been very happy.
Can you use an inductive sensor with the removable magnetic plate on the table?
Clear, concise, metrologically accurate and empirically awesome. Legend.
That test machine was brilliant! I built an i3 clone following your guide and this sort of thing is really great for developing upgrades. Thanks for being so thoughtful in the design and execution of the testing.
Thank you for this comparison! We just doing the same for a customer and we were asked this question very often as we do a lot of printer customizing.
If the tests were conducted as in the video , what do you make of Antcap's instructions regarding how the BLTouch must be mounted ?
The manufacturer's website for the BLTouch clearly states this: if the sensor is mounted horizontally it WILL give wrong results.
They don't equivocate. It is a crystal clear stipulation. It is underlined even . Would you care to comment ?
@@Traitorman.Con.14th.Sec3 except it is not just an electronic device . It is an electromechanical device and it is purpose designed to work mounted vertically . My kettle is a simple electronic device , it won't work as intended if I run it on its side or upside down.
@@Traitorman.Con.14th.Sec3 i take your point . But my inquiry is more to do with the method used to derive a conclusion , rather than the merits of this or any other device evaluated in the test. Manufacturer says this device X is designed and only works when used using method Y . Testing is done using method K , results turn out to be suboptimal . Why are the results suboptimal ? In truth we don't know ,. It may well be the product is unreliable but testing by using method K is not the way to find out. Over and out.
Yeah he should have tilted it...
If I'd design a device like this small I'd surely factor the gravity whether I'd use a spring to push the pin out or not (as it happens I'm not a flat-earther, I "believe" in gravity). Probably the pin is pulled in by a coil (with the ferromagnetic pin acting as a core). Because it needs precision, the spring, if there is one, should be weak, but the trigger point will be more influenced by the sensor orientation. I guess you could make it switch just by shaking it in the hand. To be able to ignore the gravity direction (and to have repeatability no matter the sensor orientation), the spring should be stronger, but, in this case, the precision will go away, especially with microsteppping. Factor in the uneven friction, which can be significant for a such small device and the need to keep the price small and you'll have an answer.
I don't own this small guy (BLTouch) and I didn't check the datasheet, but your observation is perfectly valid. And the BLTouch test is not valid.
Also the other mechanical sensor results aren't valid either because he's just testing the switch itself when to actually use it you'd need a mechanism to remove it to actually print.
one thing you're missing are FSRs, force sensing resistors. i use them on my delta which has a glass bed over an aluminum heat spreader.
they seem to be very repeatable but only above 0.13mm deviation, any less than that and smoothieware's autocalibration gets upset
I too use FSR's with Smoothie on my delta. I like them a lot, though they have their own drawbacks too.
Really cool video! The idea of applying statistics touched base, true engineering style, trial and error!
Thanks for your video, now I can comfortably purchase an induction sensor with the knowledge I will likely have to give up my glass bed. I appreciate all the testing you did!
Cool comparison. I was looking for a sensor to measure distance for my research project. So this helped me to understand the basics of the sensors. Thank you for that. Also is it possible to have the part list for your test setup? It will be really useful.
Great info, thanks Tom. I recently installed David Crocker’s IR sensor on my Prusa i3 box frame, which uses a 3mm glass bed and a red Mk2 PCB heated bed, inked black with a permanent marker in the sensor probing areas.
After watching your video, I tweaked Marlin so that G30 would output the trigger height to 3 decimal places (1 micron resolution). I turned on my printer and made 46 measurements with G30 and got an s.d. of 1.4 microns. I noticed a trend in the data, presumably due to temperature drift somewhere in the system (probably the steppers warming up my crappy M5 threaded rods so the trigger point appeared to get lower with time). So I did another 34 measurements and the s.d. was 0.8 microns, but still with an obvious trend.
I think these numbers are an order of magnitude better than you got with glass. Maybe it matters what is behind the glass?
I should also mention there was a thin layer of glue stick on the bed under the probing area from the last print. Also I have slightly modded my sensor board to increase the trigger height from about 2.7 mm to 3.2 mm by tweaking the angle of the outer LED using a hot-air rework station. This is so that it would clear the bulldog clips I use to hold the glass down so I could still retain my full 200x200 build area.
That's because, most likely, your bed surface is suitable for Crocker's IR sensor. The testing surfaces and setup in the video is stated as not suitable in the documentation. If you are running a glass plate on top of bare aluminum it is recommended to paint the aluminum matte black. I wish this test was done with the proper setup.
Thanks for your videos. I have experienced difficulties with sensors. After a few consecutive prints, the levelling of the bed was not accurate, but was accurate again the day after. I now make the levelling when the bed is hot, but the problem still exists. I believe that the sensor close to the bed and the head warms up little by little, and the temperature of the sensor may have some effect of the measurement.
I have a question about the inductive sensors... How much 'material' do they need below and around them and at what point does it make a difference? For instance: If I trigger right on the edge of my aluminium plate it triggers lower then when probed in the middle of my plate. Does thickness play a huge role? Do they trigger higher when there is a bearing holder right below the sensing point? How thick does the aluminium need to be to not have that influence...
You measured mostly repeatability, and mostly ignored system accuracy due to heating/cooling.
Excellent video, great charts.
Should have had an external digital DTI measure the setups actual position, logging, and dwelled (paused) after contact on the probe.
Then the DTI log would show where the moving piece actually was at probe contact, vs the start position when all was cold.
Likewise, the triggering circuit for probe hit is critical.
I did a lot of work on this, 8 years ago, on testing for lathe spindle index sensors, and we proved conclusively that sensors will give very fuzzy signals, that vary with temp/speed/luck.
A sharp triggering circuit, and a sensor tuned to give a crisp response, will be about 100x more accurate on a range of speeds vs a typical probe "hit".
I also saw that optical sensors will repeat to about 2 microns, with very simple basic cheap sensors.
Guden Thomas,
wie immer: Sehr Informativ und Professionell!
Ich danke dir für all deine Super Videos zum Thema, diese brachten mich in die Welt des 3D-Drucks!
Weiter so, alles Gute und schöne Grüße aus dem Hessen,
Otter
Would you please make a video explaining limitations of autoleveling? A lot of people seem to think it will compensate for misaligned axes (like when a dual motor Z axis loses motor sync and the X axis tilts).
Very happy with David Crocker's differential IR sensor and a sheet of Printbite.
Should've tested 3D touch as well, for example from the famous clone manufacturer Trianglelab.
3D Touch has a deviation about 3x worse than BLTouch.
I don't know where you're getting your sensors but I use 1 strip of aluminum tape under glass. Works like a charm. I used it as a retrofit onto the printrbot simple metal and it worked better than with the aluminum bed.
This is interesting. I see that microswitch w/o lever SD is about 1 um. Same as for inductive sensors. What is the point of using Bl touch or inductiv sensors if once can just put a microswitch button down on the print head and do autoleveling with it? but how to put it higher that the nozzle...
Another one to try is the piezo force sensor. Mount it in your bed support or (as I did) in your hot-end mount on the carriage and it registers contact of the nozzle to the bed. Zero probe offsets in X, Y and Z, change nozzle and all you have to do is run the auto-level. Super convenient. Precision Piezo do a good rig.
I feel like I've been in a time capsule. I got an early MakerBot CNC (#180) in 2009-ish, and was printing 3mm ABS on a raft on Kapton tape, using a .37mm layer height and no fans or heated beds. I woke up in 2021 receiving a Prusa Mini+ that prints directly on a magnetically attached and heated bed, using 50 micron layer height if I so desire, and supporting PLA, PETG and a plethora of other materials out of the box. In the old times, I tried using an optical mouse sensor for auto-leveling using focus. It actually worked with the sensor 1 mm from the bed, but I wanted something more like 10mm from the bed and never got the optics to work correctly. The idea was that the optical mouse sensor not only sensed the height, but also could read markings on the bed, or a printed sheet attached to it both for calibration precision and for communicating what to print. I never got to finish this project, but it seems like it still might be worth while: the inductive SuperPINDA doesn't even provide much headroom. Does anybody here have experience using optical focusing for Z height calibration?
For the mini-IR probe to be used on glass, the instructions say to put a black matte backing (black paper) behind it, or to paint the surface behind it matte black with high temp BBQ paint. A reflective surface behind it like aluminum will give the kind of results you saw in your tests.
Tony, see my clone of Davids mini IR, community.robo3d.com/index.php?threads/differential-ir-height-sensor-attemptin-clone-aint-sure-of-success-now-stage-10.6601/ ............Yes, you correct.....paint back of glass flat black.......Jimmy
I've been saying the same thing. This test was worst possible scenario for IR probes and IR probes only. I wish it was done properly.
newbie questions:
1) why isn't any using ultrasonic sensors ?
2) do high end printers use laser/lidar ?
thanks for the video, very informative!
1) 2) there must be distance for both of them. None of them will allow you to notice when the nozzle < 1cm. I guess there might be a workaround with offsets though when 5/10 centimeter or something like that.
Ultrasonic sensors are not precise enough to do the job.
Lidar is a process where an infared laser light creates a kind of radar around it. 3d Printers are only trying to measure distance in one direction. It is much simpliler and more practical to not use lidar. Ultrasonics are another radar type with help of a servo, Most radar like sensors are kind of inacurate but still acurate enoguh for their use cases like on a car or a plane. Where 3d printers are looking for accuracies of up to micrometers.
Always learning,
Aidan Villacampa
Prusa uses a laser to detect stoppage in their filament sensor. You can get them on Aliexpress
Are there any sensors that can map out the Z axis lead screws intolerances?
Tom, you really should continue wit the Myth-busting line. I have been experimenting with sensors I have to say that your conclusions corroborate my experience 100%. Given mechanical switches with lever , although less precise than inductive, can be used with borosilicate glass, I will stick to those. Awesome video, congrats.
This is awesome! Also fun fact: I use an aluminum mk3 heatbed with a 3/32 inch (2.38 mm) sheet of glass AND blue tape. The 8 mm inductive sensor triggers just fine for me thankfully.
Seeing as inductive probes can see through plastic/silicone, would it be sensible to put a silicone sock (or even a printed plastic cap) over the bottom of the inductive probe to reduce temperature effects? I'd be really interested to see if that helps.
Wow Tom that was a very in depth video and very well put together as usual. Thanks!
Thanks for all that testing but now I want to know which probe is best for the Scoovo X9H.
It has Kapton tape directly stuck on a glass bed with the silicon heating pad just under it.
Board: RAMBo
Firmware: Marlin
I have a unique idea for a sensor that might work better than any of these. I know it works much better than a Microswitch for sure at least....when I built it initially I didn't know what use it may be, but now it is all quite clear! Thank you Thomas! You inspire me!
Hello. Where can we read the results please? I am very interested by mecanical switch results (with and without lever)
As I understand, at 7:24; mecanical without lever is more precise than BLTouch ? cool!!
"Alumininium" alone earned my subscription, whether it was intentional or just a typo. :D This was very informative, thanks for taking the time to do all the testing!
The entire world uses "aluminium." Only america uses "Aluminum."
John Z Watch the video again if you didn't see what I was commenting on. I promise that no matter what country you're from, it's not generally spelled that way. Tom inserted a small item into the video that is internationally known as a "joke", and is intended to provoke a humor response in humans with at least average spelling and reading skills.
+John Z Seriously dude, look at what I wrote in my original comment,
re-watch the video, and then try to be amazed at how dense you actually
are. The word is intentionally misspelled both by Tom in the video, and
by me above. There are extra letters in it. It doesn't say "aluminum" or
"aluminium". In your efforts to claim Internet Superiority, you've
completely missed the obvious.
I've just discovered your channel and love the friendly sound of your voice. I want to do a few small projects and I was weighing paying for prints vs buying my own printer. I found the new matter mod-t for 299 but wondered was there something better for beginners that maybe will help them as they transition to an intermediate level of skill without going a lot higher ( +$200 US) in price. In any case, I enjoy your videos. Thank you.
Great, informative videos. Thank you. I was wondering if you tested the BLTouch horizontally or vertically? I think that its accuracy is affected by gravity and would benefit from a vertical test. I agree that it is a good alternative for glass probing.
A coil-operated pin this small and light should not be influenced by gravity at all. The testing method is perfectly acceptable in this case.
@@pellin-unleashthebiker3101 not when you're dealing with micron-level precision. I calibrate measurement equipment to this level as part of my job, and misusing equipment like this is a total non-starter if I want something to perform accurately.
@@TROPtastic did you look at the precision of the bltouch horizontally in this video? It is only some microns, so still perfect. So the conclusion is it really does not matter in this case. For something else it may, but not for a simple 3d printer.
Very informative video that highlights the differences in sensor types. Just what I was looking for to help decide what sensors to look at.
Tom, if you retest the optical sensors, I would like to see it done with different color materials behind the glass (specifically black, white, red and peach- the common colors of heaters). Would also like to see how ambient IR light affects the repeatability, i.e. fluorescent lighting and sunlight. I had very mixed results with David Crocker's IR sensor during probing with sunlight coming in through the windows, for example, vs. probing in a dark room.
The IR sensor I make doesn't like bright sunlight (no clouds in front of the sun) shining directly on the bed and reflecting into the sensor. The current version 1.2 will flash the LED rapidly to warn you when it detects that. Otherwise, it isn't affected by ambient light. Incandescent light - especially halogen - affects some other types of IR sensor, but not mine.
I'd also like to know what material was behind the glass when Tom did his tests. The reason that glass and other transparent materials can cause problems for optical sensors is that you get 2 or 3 reflections: one from the front, one from the back, and possibly a third from whatever is behind the glass. We want the one from the front to be stronger than the others, which is why we recommend a black surface behind the glass.
I liked this video a lot, I simply wish the sensor I use (David Crockers IR Sensor) had a slightly deeper/more in depth review. I am given more confidence to continue using it, however, will no longer calibrate it with the heated bed on.
That is a variable I had not even thought twice of! I simply set the bed for 55° for my PolyMax PLA and calibrated it when it was done heating right before the print...Thank You for that!
Your videos are all excellent, but this one is superb! We spent so much effort learning this on our own... at Kikai Labs we use a capacitive sensor, take 16 measurements prior to starting each print job and level the bed automatically. It works great. We do calibrate the sensor first using three different bed temperatures, as it does matter as you pointed out. Interesting observation about the humidity, though... I used capacitive so that it would measure the glass surface, and we use a mirror instead of plain glass, and get better results. Also we found that the smallest the measuring distance, the more accurate it is, so we measure from 2mm away using a 12mm sensor.
Y que le paso a Kikai Labs?
I'd be interested in your review of the piezo electric z level sensor, it looks super interesting to me. It has a bed option and a hotend option, it's always there and has no XY offsets.
If you ever re-run this test I'd love to see the piezo sensors be included. I'm about to give one a try myself but curious to see your experience vs mine/others
Almost 2 years later now and I still adore my BLTouch. Even if it statistically or scientifically isn't the most accurate out there, as you say, it's 'more than good enough'. It's so rare to have a calibration fail.
Usually there is a spec sheet prescribing optimal materials and thickness with respect to frequency (typically a few mils for ferrous and 2-10x thicker for nonferrous). The relevant parameters are permeability and resistivity of the material. Ferrous materials and sensors generally can give higher resolution with larger sensors, and non-ferrous gives better resolution with very small sensors (even though typically smaller sensors give a shorter range), but there are universal sensors that have an equal correction factor for either.
Generally ferrous sensors would probably work best in this application, and Generally target size should be 1-3 times the sensor diameter. Also, in terms of layered copper/alum tape, remember that you are isolating the layers electrically, so the induced eddy currents will be smaller (just like transformer laminations)… meaning you may need more layers of tape than you would expect.
If I were adding a proximity sensor after the fact, I would get a medium-distance ferrous sensor, tape ~1mm thick steel disks to the bottom of the bed at the sensor points, fiddle with the distance, and call it a day.
This is amazing! Finally a useful application of the scientific method to get some hard data. It is so much more convincing than people's opinions. Love your videos!
How do i use a microswitch for the auto leveling? I mean where should i put it, so it is not interfering with the nozzle and mess up the print?
It seems a good option for me. I have glass bed on top of an aluminum sheet... (DIY)
Wow love the detail and different perspectives covered in this video. I'm a sucker for this sort of data. If inductive sensors have their detection range halved, and there is a sheet of window glass (2.38mm thick) on top of the bed, doesn't that mean that both 2mm and 4mm inductive sensors are right out?
I'd love to print ABS right on the aluminum but nothing sticks to it without a carefully maintained pile of glop (glue/slurry), where glass works for weeks with a single light hairspray application.
How does Marlin deal with the measurements from the probe? Will it compute a plane with minimum distance to each point, or does it compute a non planar surface to deal with deformations of the bed or bending/wear of the linear supports carrying the print head.
Very, very nice testing here! Thanks Thomas!
I've been using a little microswitch (with lever) on a metal strip moved by a servo. It's working flawlessly for years now.
(all metal printer, no 3d printed parts)
Nice to see that this turns out to be a precise way indeed :-)
Hi, Are you still using the micro switch?
love the video. can you publish the accuracy finding?
Thank you very much for this very interesting & informative video!! I love those technical contributions.
Wow great video thanks. This should get a lot more attention
I like to print on glass and don't have auto leveling at the moment. The micro switch performed well but could you test it with a servo attached and see how it performs.
Informative as always. I use a LJ12A-4-Z/BY with an LM7805 VR on the output this gives a steady 5 volt trigger signal.
If you increase the input voltage does it does it strengthen the sensing distance.
Necro, but thanks a billion for making this video. Would be interested to see what new sensors are out there, and learning more about the LIDAR from the Bambu Lab X1C
Great Video - seriously stellar quality and effort clearly went into this. The pinned top comment is exactly the question one is left wondering. Great job!
out of all the sensors that you tested what one would you recommend to use on an aluminum bed with 3-4mm glass
i guess you can't go wrong with the bl touch if you print on glass
Why do you need a leveling sensor on glass aren't they flat?
@@chloemcholoe3280 The glass (print bed) will be a flat piece of glass. However, it is then mounted inside a mechanical device. If one edge is mounted lower than the opposite edge then your first layer of 3d printing will not adhere well to the glass. The print bed must be leveled to ensure that the 3d print head glides over the whole print bed at the same height before a 3d print begins. Then the 3d printer will reliably 3d print the first layer and subsequent layers. This leveling can be done manually, with no automated sensors, or leveling can be done with automation using the kinds of sensors tested.
@@paull1316 yeah. my point was the reason I for these probes in my opinion is for fixing warpage. with glass you can just like live adjust the 4 corners as it's printing and it takes a few seconds really
@@chloemcholoe3280 - glass can be warped or bowed if not made correclty. Also, glass isn't the only height variable. its possible that the gantry rail that the hot end rides on can be bowed.. which may change the height of the nozzle as it moves back and forth.
This was awesome. I love these technical kinds of videos. Thanks for the info, Tom!
Incredibly thorough, concise review! Keep up the great work
Invaluable advice and comparisons Tom. Massive appreciation mate.
When you tested the inductive sensors you didn't take in account the heterogeneity of an aluminium bed.
While printing with my VORON I have experienced inconsistencies with inductive sensors and my PEI-painted Aluminium bed from Clever3D. Basically the trigger distance was consistent on the same spot (just like in your tests), but varied when probing a different spot... and you need to probe at least 3 distant spots.
Today I switched to the VORON 1.5 X carriage, which uses the hotend itself (mounted on a mechanical microswitch) as a probe and the situation is vastly improved because a microswitch isn't affected by materials and temperature.
EDIT: In theory this applies to IR probes too, as reflectivity may vary across the bed.
Hello Tom, I have a aluminum heated bed and a genuine e3dv6, and I was trying to use the autobed leveling making a "button" that closes when the brass nozzle touches the aluminum bed. Turns out that there is some kind of "coat" or resin or something in the surface of my heated bed so it doesn't conducts eletricity. Do you know what is this coating and why? And is this a good idea, use the nozzle itself?
Any chance to give me an idea, how to set up autoleveling in smoothieware, how to level the bed and what gcode should I put in start/end code? I have a 18mm capacitance proximity sensor on hotend carriage.
Very informative. I was wondering if you could repeat this experiment with the new MINDA probe from Prusa Mini. Many folks seem to be having a problem with leveling because of the variations between cold and heated bed.
Awesome test! Very methodical tests. I have the cap sense and have noticed the heat issues. I think I'll switch to inductive.
I would like to try a glass bed over the aluminum lower plate. I wonder if the the longer range inductive sensor would be practical at detecting the aluminum glass combo.
I'm wondering if copper sheet might work better than aluminium foil for inductive sensors on a glass bed.
Good review, thanks.
How do you feel about auto levelling? I personally don't like the idea of the Z axis having to move during a layer print. If I'm worried I just use a raft, which will end up levelling before the actual model prints. What I really want to see is 3 servos to level the bed corners.
Servos are just so unstable. What you want is the thumbscrew on top and having a hook for the nozzle to turn it. *g*
HenryLoenwind no idea what crazy setup you're envisioning but I guess it's not the one I was... I use servo in its loosest term: any motor hooked up to a potentiometer to increase positional accuracy, not as in your hobby king swivel arm 180° ones. As such, a small motor attached to the thumbscrew so it works as a linear actuator and a linear potentiometer hooked to that setup for positional feedback. Doesn't seem too ludicrous.
You were suggesting using the nozzle to turn a thumbscrew? That grates against my engineering sensibilities quite badly.
+HenryLoenwind
Oh man, the mental image you just gave me of a printer spinning its nozzle around a hooked screw to 'adjust' itself before it feels comfortable enough to start producing a part is hilarious to me, for some reason.
Guess you missed the G at the end. He was joking. LOL!
I thought it particularly funny a guy with "Mirth" in his name would miss the joke. :-D
My team is actually working on a printer for the market with all best stuff available, and that includes 3 steppers for Z axis for REAL bed leveling, not some sort of lame workaround. ;) Fast hotbed (0-110C in 25 seconds), 300x300x450 build volume. Won't be cheap though. Good stuff is never cheap :)
I thought you needed an PNP sensor? (at least for a ender 3) Can you use either and set the high/low trigger in marlin firmware?
Also, what voltage divider did you use?
Have you try the membrane film based zprobe (basically just like a membrane keyboard/contact switch)? I find it is the best zprobe for delta printer as it can be easily mounted directly under the nozzle to give the best on-points reading that a delta auto calibration needs (no zprobe offsets), without being affected by heavy trigger force that could potentially deform the belts/pivot points etc.
I have a Taz5. Would love to see what is the best option and how to install it.
Very informative, thats exactly the content I prefere. Some more mythbusteresque explosions would be neat.
Excellent :)
Yeah, let's slam that probe through that glass with some speed! *g*
great video! cool rig, shows the power of rapid prototyping. been thinking about Auto leveling, I have a delta that uses pressure pads, would be great to see how well it stacks up, down side is Auto leveling is tapping printer head into the bed every time.
Great treatment of these of bed sensors. Could you the same to these interfaces options
My apologies if this was asked already, but how does the touch sensing of the lulzbot taz6 compare?
for my scratch built reprap delta we used a simple conductive type, running 1 wire to the nozzle, and another to the aluminium bed, reeally simple and easy, and no need to worry about effector tilt
You should try the aluminum tape trick with the optical sensors, I have a feeling the Sharp will benefit from it.
Also, try doing tests with multiple sensors, like one for each corner of a bed vs a single sensor.
im using a glass bed and some aluminium foil in the corners underneath it, it manages to sense it but is it a good idea to continue using this construction?
I would love testing of a few more sensor options: FSR, seemeCNC accelerometer and piezo-electric disk
David Crocker designed his sensor to use black paper backed glass, however I don't see any test of that case.
Can you redo the tests but use the BLT touch the way it supposed to be vertically mounted not horizontally
What would the point be? He flat out told you it was done sideways to remove backlash from the equation and limit the results to just the component . Turning them upright wont change how the devices reacted mechanically one bit .
@@wind5250 and apparently you didn't read the instructions for the bill touch it supposed to be vertically standing not horizontally to work properlyso if you want to test something the way it was made to be tested and used that's the way you do it not for non-science a backlash there's no backlash when there's up and down
@@mathewphillips4185 not only did i read the instructions i own one. Laying it on it's side is just as valid as standing it up. The fact his results are within spec as well as repeatable , and comparable to other reviews more than demonstrates this .
@@wind5250 you read the instructions but apparently didn't follow them (and are misusing the equipment you own, but that's your choice). It's simple physics to demonstrate that there is a mechanical difference in how the probe behaves when it is vertical vs. horizontal:
When vertical, gravity acts on the spring to pull it down and keep it in its rest position. When the probe hits the surface, it can cleanly move upward until it can trigger its Hall sensor. When horizontal, not only is gravity no longer aligning the spring correctly, but there is additional friction resisting the movement of the probe (thus causing the probe to move fractionally less for the same amount of deflection). When the probe contacts a surface horizontally, there is no force preventing the probe from deflecting sideways ten+ microns before triggering the Hall sensor, leading to a loss of precision.
If using your equipment like this works for you, that's great, but it's objectively incorrect to say that orientation doesn't have an effect on the BLTouch and other precision equipment.
@@TROPtastic You took all that time to write a reply but obviously not enough to actually comprehend before hand . First of all i never said how i mounted the device let alone i mounted it sideways. What i said was it would make no difference to repeat the test upright because the results are within spec and repeatable which they are .
You are literally arguing that the device will perform better than the manufactures tell you it will by turning it right side up.
Anyone who actually uses this device knows they give invalid readings after time the only valid questions would be what version this is and was this one new or used .
Hi,
I 'm not sure what's happening with my sensor which is the 8 mm inductive npn one. Before adding the Voltage divider I put in 12V across blue and brown and get a 12V signal on the black when I hold the sensor near metal and 0V when the metal is removed. When I add the Voltage divider as Tom did (double checked and resistors measured with multimeter to test) I get 3.6V signal on black when not near metal and 0V when I put sensor near metal.
What have I done wrong? can I use this as a normally closed (nc) switch for bed levelling/tramming?
would it not be best to combine different sensor types in one device, for example inductive and contact, and measure the deviation of these for error analyses?
would you think it's worth it for capacitive sensor and active dehumidifiers? or maybe a butt load of those silica packets
Would the inductive sensors work better on the magnetic beds, or indeed with magnetic tape under the bed in the positions you are probing?
any chance we could see a version of this test with some piezo action too?
Great video. Love the rig you created. Is there a normal term for the slide you used that has to rails built in? I am thinking of using one on a 3D Printer I want to build from scratch
I have not seen any video like this one, wow! super informative! keep it up man
Hi, I use some lj12a3-4-z/bx on my printers and it triger on a glass plate with some almuminium foil sticked with rubber tape on it. Also supplied in 5V directly plug on my borads without any led or resistor.